Gastrointestinal Stromal Tumors (GISTs)

Published on 20/05/2015 by admin

Filed under Gastroenterology and Hepatology

Last modified 20/05/2015

Print this page

This article have been viewed 2715 times

CHAPTER 30 Gastrointestinal Stromal Tumors (GISTs)

Recent research advances have greatly elucidated the pathogenesis of gastrointestinal stromal tumors (GISTs) at the molecular level, and translation of this understanding into highly effective, molecularly targeted therapies has followed in rapid succession. GISTs has evolved rapidly from an arcane disease known only to a few investigative teams worldwide into a proof of concept model for cancer medicine, personalizing highly selective therapies to the dysregulated cell signaling that contributes to the abnormal behavior of these neoplastic cells.

INCIDENCE

It is difficult to obtain accurate data regarding the true incidence of GISTs. This is because of referral bias, which concentrates GIST cases with a worse prognosis and a more malignant behavior in academic cancer centers, and also the lack of definitive diagnostic techniques before the molecular definitions of GISTs in 1998 and later. Before 2000, the number of new GIST cases in the United States had been underestimated and underreported. However, it is now recognized that the true incidence of GISTs is much higher than previously believed. Much of this is to the result of an increased recognition of GISTs; new molecularly targeted drugs, such as imatinib mesylate (Gleevec), have become available. A population-based study to assess the incidence of GISTs using the most up-to-date criteria has estimated an annual incidence of approximately 15 cases/million.¹ This has increased the estimated incidence of GISTs in the United States alone to approximately 5000 new cases/year.²^,³ However, even this figure may underestimate the actual incidence because many smaller lesions, including those detected incidentally by many gastroenterologists on endoscopy performed for other reasons, are almost certainly not revealed in these incidence figures. Not all GIST cases will prove to be life-threatening, because many patients who present with GISTs have limited disease with lesions that are small (smaller than 1 cm) and fully curable with optimal surgery as first-line therapy. It remains a question as to whether there are any true GIST lesions that might be able to be observed without resection.

GISTs occur predominantly in adults, with a median incidence in the fifth decade of life, but these tumors can occur across the age spectrum, from infants to older adults. The incidence has been reported to be slightly higher in men than women. Recent studies of surgical or autopsy specimens of stomach samples resected for non-neoplastic disease have documented a remarkably high incidence of occult microscopic GIST lesions, in the range of 20% to 35%.⁴^,⁵

LOCATION

Most GISTs (60% to 70%) arise in the stomach, 20% to 30% originate in the small intestine, and less than 10% in the esophagus, colon, and rectum (see later). GISTs can also occur in extraintestinal sites in the abdomen or pelvis such as the omentum, mesentery, or retroperitoneum.⁶^–⁸ However, data on sites of GIST origin are somewhat limited. Large studies before 1999 may have included other subtypes of mesenchymal neoplasms because these series were not selected by immunohistochemical or molecular genetic markers specific for true GISTs.

CLINICAL FEATURES

The clinical presentation of patients with GISTs can vary tremendously on the basis of the anatomic location of the primary lesion, as well as other factors, such as tumor size and presence or absence of symptomatic metastases. For most GIST patients, the initial detection of GISTs may be an incidental finding or result from evaluation of nonspecific symptoms. Symptoms from GISTs are noted usually only after tumors have reached a large size or have impinged on a specific anatomic region (e.g., constricting gastric outflow). Most symptomatic patients present with tumors larger than 5 cm in maximum dimension. Symptoms at presentation may include a palpable abdominal mass or swelling, abdominal pain, nausea, vomiting, anorexia, and early satiety. Although it has been reported that up to 40% of GIST patients present with acute hemorrhage into the gastrointestinal (GI) tract or peritoneal cavity from tumor rupture, it is clear that such reports are dependent on referral bias of patients with large GIST lesions or multifocal disease.

The vast majority of GIST metastases at presentation are intra-abdominal, with metastases to the liver, omentum, or peritoneal cavity.⁶ Metastatic spread to lymph nodes and to other regions via lymphatics is rare; most lesions thought to be nodal metastases by imaging studies simply represent metastatic deposits of tumor nodules in the omentum or peritoneum rather than true lymphatic spread of the disease.

PATHOLOGY AND MOLECULAR PATHOBIOLOGY

UNIQUE CLINICOPATHOLOGIC SUBTYPE OF SARCOMAS (MESENCHYMAL NEOPLASIA)

Although relatively rare compared with epithelial cancers, GISTs represent a form of sarcoma that comprises approximately 1% to 3% of all malignant GI tumors. As noted, the diagnostic criteria for GISTs before 1999 were somewhat subjective, controversial, and perhaps even a bit confusing, with a great deal of room for interobserver variation. The term GIST was initially a purely descriptive term applied by Mazur and Clark in 1983 to define intra-abdominal tumors that were not carcinomas (i.e., nonepithelial) and that did not exhibit features of smooth muscle or nerve cells.⁹ The morphology of the cells was the dominant feature driving the diagnostic scheme. However, pathologists subsequently recognized that there was no completely clear differential expression of muscle or nerve antigenic markers when careful immunohistochemical analyses were performed on samples identified as GIST lesions by cellular morphology. The expression of differentiation antigens used as markers for muscle cells (e.g., smooth muscle actin) and nerve cells (e.g., S100) was noted to vary widely in GI mesenchymal lesions, leading to interesting hypotheses about whether GIST lesions from different patients were attempting to recapitulate distinct myogenic or neural programs of differentiation.

To accommodate these empirical observations, it was proposed that approximately one third of GIST lesions differentiated along smooth muscle lineages, another third were neurogenic in origin, and the final third lacked any detectable lineage-specific markers (null phenotype) by immunohistochemical analysis.¹⁰^–¹²

Nomenclature for GISTs had also been confusingly complex before the availability of molecular- and mechanism-based diagnostic tests. Because of limited understanding of the disease before 1999, GISTs were most commonly diagnosed as leiomyomas or leiomyosarcomas because of the histologic resemblance to these smooth muscle neoplasms. Other terms that had often been applied to a GIST included benign leiomyoblastoma and, recognizing some of the neural characteristics, plexosarcomas¹³ or gastrointestinal autonomic nerve tumors (GANTs).¹⁴ All these terms are now recognized as referring to GISTs. Insightful studies by several pathology groups have noted that the panoply of tumors lumped together as smooth muscle tumors of the GI tract were likely not simply leiomyosarcomas nor benign leiomyomas; a subset of these tumors originating in the bowel wall had several unique histologic features, probably representing a totally different diagnostic group altogether.¹⁵^,¹⁶ Additionally, clinical oncologists had noted that putative leiomyosarcomas of the GI tract had a completely different prognosis than true leiomyosarcomas arising in other parts of the body (e.g., the uterus). The leiomyosarcomas of the GI tract were extraordinarily resistant to standard chemotherapy regimens, further supporting the notion that these lesions represent a different form of cancer entirely.

Immunohistochemical analysis of GISTs in the early 1990s attempted to find specific markers that might distinguish GISTs from other spindle cell tumors of the GI tract, such as schwannomas and sarcomatoid carcinomas. There was some initial enthusiasm for the CD34 antigen as such a marker; however, this antigen is also expressed by hematopoietic stem cells and by vascular and myofibroblastic cells. Also, the sensitivity and specificity of CD34 are low because only approximately half of GIST cases express CD34, and other smooth muscle, myofibroblastic (e.g., desmoid), or Schwann cell tumors can also express CD34. Therefore, CD34 is not a reliable marker to distinguish true GISTs from other neoplasms.¹⁷^,¹⁸

It is clear that before approximately 1999, there were no objective, reproducible, and clearly defined criteria for the diagnosis and classification of GISTs, and it is likely that several types of epithelioid and spindle cell tumors were included in the clinical diagnostic category of GISTs. Similarly, many true GIST cases had various different diagnostic labels, such as leiomyoblastomas and GANTs. This makes the interpretation of published GIST series accumulated before the year 2000 difficult, given the heterogeneity that was reflected by the diagnostic term GIST before the widespread use of specific kinase-directed diagnostic and molecular markers for this disease.

UNCONTROLLED KINASE ACTIVATION: PRIMARY MOLECULAR PATHOGENESIS

A critical advance in the understanding of GISTs at a molecular level occurred in the late 1990s, with the recognition that these tumors exhibited some histopathologic similarities with the pacemaker cells of the gut known as the interstitial cells of Cajal (ICCs).¹⁹ ICCs are normally present in the myenteric plexus and serve to coordinate gut peristalsis by assisting the linkage of smooth muscle cells of the bowel wall with the autonomic nervous system (see Chapters 96 to 98). GIST cells and ICCs have certain ultrastructural features in common, such as the combination of neural and muscle phenotypes.

The molecular pathogenesis of GISTs was advanced further by a key observation made by Hirota and colleagues ²⁰ in 1998. This group was studying the role played by the KIT receptor tyrosine kinase (RTK) in cell growth and development. Expression of the KIT RTK can be detected by immunohistochemical staining for the CD117 antigen as a marker of the KIT protein. In normal cell signaling, KIT binds its ligand, known as stem cell factor (SCF) or Steel factor; ligand binding brings together two molecules of KIT with subsequent activation of a signaling cascade. A homodimeric complex is formed with these two KIT receptors, leading to cross-phosphorylation of critical tyrosine residues in the intracellular domains of KIT, which activate signal transduction pathways downstream of KIT. The net physiologic effect of normal ligand-induced KIT activation is the controlled stimulation of cell proliferation and enhanced cell survival; therefore, uncontrolled activation could theoretically lead to neoplastic growth and transformation of cells.

Hirota and colleagues ²⁰ recognized this potential mechanism and provided the critical confirmation of this theory at the cellular and molecular levels. This same team, in a project led by Nishida, also expanded these observations to familial GISTs, with germline carriage of an activating mutation in the KIT gene encoding the RTK.²¹ This elegant work supported some of the key biologic similarities between GIST and ICC cells²²^–²⁵ because both cell types had been shown to express the KIT RTK. It is likely that the cells of GISTs and normal ICCs share a common precursor cell.²⁶^,²⁷ The KIT RTK and SCF play essential roles in the development and maintenance of normal ICCs, as well as other cells, including melanocytes, erythrocytes, germ cells, and mast cells. KIT expression is noted in the vast majority (>95%) of GISTs, but KIT is not expressed by true smooth muscle tumors of the GI tract nor by stromal tumors at other anatomic locations, such as endometrial stromal tumors. Although the origin of the neoplastic cells of GISTs remains a matter of active investigation, some data suggest that GISTs originate from CD34-positive stem cells residing within the wall of the gut, which can then differentiate incompletely toward the ICC phenotype.²⁷^–²⁹

GISTs characteristically exhibit expression of CD117 by immunohistochemical assays (>95% of lesions but, importantly, not 100%, because there are true KIT-negative GIST cases).³⁰ Levels of expression of CD117 (KIT) are generally diffuse and strong in the spindle cell GIST subtype (Fig. 30-1). In contrast, in the epithelioid subtype, CD117 expression is typically focal and weakly positive in a dot-like pattern (Fig. 30-2). As noted, CD34 expression is neither sensitive nor specific for GIST because this antigen can also be noted in desmoid tumors, and approximately 30% to 40% of GIST lesions are negative for CD34.²¹^,²⁶ True leiomyosarcomas express two smooth muscle markers, smooth muscle actin (SMA) and desmin, but fail to express CD117. Schwannomas are usually positive for the neural antigen S100 but are also negative for CD117. Normal mast cells and ICCs in the surrounding stromal tissues serve as ideal positive internal controls because these normal cells strongly express CD117. Activating mutations in the KIT gene were identified in five of six cases of human GISTs originally analyzed by Hirota and colleagues,²⁰ with evidence that the mutations resulted in uncontrolled, ligand-independent activation of the KIT kinase. Genetically engineered cells harboring the mutant overactive KIT proteins were tumorigenic in nude mice, serving as proof of concept that the malignant phenotype was directly induced by the aberrant signaling pathways associated with uncontrolled KIT kinase activation.

Figure 30-1. A, High-power photomicrograph of a typical spindle cell gastrointestinal stromal tumor (GIST). The cells are monomorphic, have abundant pale, eosinophilic, fibrillary cytoplasm, and lack mitotic activity. B, KIT immunostaining. This medium-power photomicrograph of a spindle cell GIST shows diffuse and strong cytoplasmic immunoreactivity for KIT. The entrapped muscle fibers from the bowel wall are negative by CD117 immunostaining for KIT (A, Hematoxylin and eosin, ×100; B, CD117 immunostain, ×100).

(Courtesy of Dr. Brian P. Rubin, Cleveland, Ohio.)

Figure 30-2. High-power view of an epithelioid gastrointestinal stromal tumor (GIST) showing epithelioid cytomorphology, fibrillary cytoplasm, and a lack of mitotic activity (Hematoxylin and eosin, ×200).

(Courtesy of Dr. Brian P. Rubin, Cleveland, Ohio.)

The oncogenic potential of mutant, uncontrollably active KIT in the pathogenesis of GISTs in humans has also been supported by the identification of familial syndromes with an autosomal dominant inheritance pattern and an abnormally high incidence of GISTs, usually occurring as multiple foci within any affected individual (see later).²¹^,³¹^,³² Genetic analysis of such kindreds reveals that they harbor germline-activating KIT mutations, similar to the mutations that were first described in sporadic cases of GISTs.

With analyses using optimal tumor specimens and sophisticated technology, it has become clear that KIT mutations can be detected in more than 90% of GIST cells.³³^–³⁵ Constitutive activation of the KIT kinase enzymatic function has been reported to characterize every GIST sample analyzed by immunoblotting technique, even in cases in which there are no detectable mutations in the KIT gene. The mechanisms whereby nonmutated (wild-type) KIT is maintained in an uncontrollably phosphorylated state are poorly understood and are likely to be fertile ground for future research, with therapeutic potential.

Importantly, the vast majority of GIST cells at initial presentation demonstrate only a single site of mutation in the KIT gene; complex genetic changes in untreated GISTs at initial presentation are vanishingly rare. Gain of function mutations have been identified most commonly in exon 11 of KIT (up to 70% of cases), an exon that encodes the intracellular juxtamembrane domain of the KIT protein. Mutations in the KIT gene locus have also been described in other regions including (in decreasing order of prevalence) exon 9 (the KIT extracellular domain), exon 13 (kinase domain), and exon 17 (kinase domain).³³^–³⁵ Structural biology studies have revealed the mechanism whereby normal (wild-type) KIT is kept in an autoinhibited conformation until ligand binds; mutational changes in conformation interfere with this autoinhibition and lead to a structural basis for the aberrant activation of the KIT kinase function.³⁶^,³⁷

Another key advance in the understanding of GISTs has been the recognition that signaling through other uncontrolled kinases in addition to KIT could drive the neoplastic phenotype of GIST cells. Specifically, it is now recognized that approximately 5% of GIST cells are not through activation and aberrant signaling of the KIT receptor, but rather through mutational activation of the structurally related kinase known as the platelet-derived growth factor receptor-alpha (PDGFRA).³⁸^,³⁹

The definitive diagnostic criteria of uncommon CD117-negative lesions that are nevertheless truly GISTs are currently somewhat obscure. GIST lesions can be heterogeneous in the expression of CD117, even within a single mass. It is possible that a needle biopsy could yield cells consistent with a GIST yet be CD117-negative simply by sampling bias alone. However, expert pathologists can also define a rare subset of GISTs (<5% of cases overall) that have no CD117 expression; these are most likely dependent on an alternative kinase such as PDGFRA.³⁸ Molecular analyses of the KIT and PDGFRA genotypes may be useful to define with certainty the group of rare patients with CD117-negative GISTs in the future. Now, it is important to note that the diagnosis of a GIST should be made on the grounds of morphologic, clinicopathologic, and immunohistochemical data, as well as possibly molecular analysis if there is any ambiguity from the other pathologic assessments.

KIT mutations have also been documented in small GISTs (<1 cm in greatest dimension)⁴⁰; such lesions are most often detected incidentally (e.g., during upper endoscopy for reflux symptoms) and may appear morphologically benign. These findings support the hypothesis that activating mutations in the KIT proto-oncogene represent an early event in the transformation from a normal precursor cell into a GIST lesion. Because lesions in familial GISTs (see later) may not present clinically until the second or third decade of life, or even much later, it is likely that second hits are necessary to attain a more aggressive malignant phenotype. The other key signaling steps that confer a more malignant phenotype to GIST cells remain obscure. However, the unique aspects of the signaling cascades in GISTs are being actively elucidated, and these appear to differ from KIT signaling in hematologic cancers. For example, the STAT5 pathway of leukemic cells is not typically activated in GISTs, whereas STAT1 and STAT3 are activated at a high level.³³

BENIGN GASTROINTESTINAL STROMAL TUMORS

The literature before 2000 was somewhat confusing about whether mutational status of KIT could distinguish between so-called benign and malignant GISTs. With the recognition that KIT mutations can be found in even the smallest GIST,⁴⁰ there is now consensus that KIT genotype alone cannot account for differences between GISTs that may behave in an indolent manner (and which, when small, may be curable by optimal resection alone) versus those that are clearly aggressive and malignant by all functional definitions. It is important to note that a well-differentiated benign cell morphology alone should not provide any reassurance that an individual GIST lesion will pursue a benign clinical course.

Consensus was reached at a meeting held at the National Cancer Institute (NCI) among pathologists with expertise in GISTs. This consensus defined the two most reliable prognostic factors for behavior of a primary GIST as the size of the primary tumor and the number of mitoses, reflecting the proliferative activity of the cells.² Other factors, such as the specific histologic subtype (epithelioid vs. spindle cell; see Figs. 30-1A and 30-2), the degree of cellular pleomorphism, and patient age may have some contribution to prognosis but are most likely to play a minor role in determining the clinical outcome. Recurrence and survival rates have also been reported to correlate with the location of the primary GIST lesion, with small bowel tumors showing a somewhat worse prognosis. Most important, the consensus panel emphasized that no GIST lesion should ever be considered completely benign. This is because morphologically benign lesions have a finite capacity to recur and metastasize, occasionally several years after initial presentation of the primary disease. Nonetheless, further research is necessary to define the true risks of tiny GIST lesions (<1 cm) because these are far more common than previously appreciated.

DIAGNOSIS

The revolutionary changes in the scientific understanding, diagnostic evaluation, and therapeutic management of GIST patients (discussed later) has led several professional organizations to develop consensus-based (and, whenever possible, evidence-based) clinical practice guidelines. The National Comprehensive Cancer Network (NCCN) has developed extensive publically accessible guidelines to assist clinicians in the management of GIST patients (see the GIST and Soft Tissue Sarcoma Guidelines on www.nccn.org). Also, the European Society of Medical Oncology (ESMO) has published expert-driven clinical practice guidelines.⁴¹ Given the rapid progress in this field, these guidelines are reviewed at least annually to ensure that current information is available to assist clinicians in providing the most up-to-date care and accurate information to patients.

The diagnostic evaluation of a suspected or proven GIST is similar to that of other GI neoplasms. The most important element is to keep GISTs in the suspected differential diagnosis of any mass lesion noted throughout the length of the GI tract, as well as in extraintestinal sites of the abdomen and pelvis. As for any GI evaluation, the site of lesion may determine which diagnostic tools are most appropriate for the patient.

DIAGNOSTIC APPROACH TO STROMAL TUMORS BY SITE

Esophageal Tumors

GIST lesions may arise within the esophagus, although this is a rare presentation for larger lesions. Most esophageal GIST lesions are noted incidentally during upper endoscopy performed for some other unrelated symptom or disorder, such as reflux esophagitis. Esophageal GIST lesions may be small (only a few millimeters in size) in this location and may be resected using endoscopic techniques.⁴² Margins may be involved if a lesion, unsuspected as a GIST and thought to be benign, is simply popped out using an endoscopic procedure. It remains unclear whether watchful waiting with serial endoscopic follow-up is appropriate for any patient with small GIST lesions (<1 cm in maximal dimension).

As noted, histopathology showing putatively benign GIST cells cannot be viewed reassuringly because histology does not perfectly predict the malignant behavior of GISTs. A careful risk-based assessment that takes into account other aspects of the tumor, as well as patient-specific factors (e.g., age, comorbidities, patient preferences), must be performed. It will be important for prospective research to accumulate a larger objective database regarding the outcomes of such patients so that medical decision making can be based on solid evidence. Although a small esophageal GIST lesion is probably a single primary, it is appropriate to take a careful family history and perform computed tomography (CT) scanning of the abdomen and pelvis to ensure that no other lesions are present in the patient’s family, especially if the disease is first detected in a young person. Some series have attempted to distinguish between clinical outcomes of patients with GISTs and those with leiomyosarcomas or other mesenchymal neoplasms, such as leiomyomas.⁴³

Gastric Tumors

The most common primary site for GISTs is the stomach. Most GIST lesions are submucosal, rather than mucosal, without overlying ulceration (Fig. 30-3). This explains why many GIST masses may only be visualized on endoscopy as a subtle, smooth protrusion with overly normal mucosa.⁴⁴ Also, this submucosal localization can make diagnostic biopsy through an endoscope difficult. It is not uncommon for superficial biopsies to reveal only normal mucosa, whereas deeper biopsies or histopathology from a definitive resection would show the true underlying GIST cells.

Figure 30-3. Spindle cell gastrointestinal stromal tumor (GIST) of the stomach. The lesion is well circumscribed and does not invade the muscularis mucosa. Invasion of the muscularis mucosa is considered an adverse prognostic factor (Hematoxylin and eosin, ×50).

(Courtesy of Dr. Brian P. Rubin, Cleveland, Ohio.)

Small Intestinal Tumors

The second most common site for GISTs to arise is in the small intestine. These lesions have been reported to have a worse prognosis overall, for unclear reasons. One possibility is that small intestinal lesions simply remain asymptomatic for a longer time before the disease is brought to medical attention and detected. If the differences were solely because of this lead time bias, small intestinal GISTs should present with significantly larger lesions than other primary sites, which is often the case. GIST masses in the small intestine tend to arise in the jejunum, followed by the ileum and the duodenum. The large lesions may be highly vascularized and present significant risks of bleeding, even with only a biopsy.

Because complete surgical resection is the treatment of choice for localized GISTs, there is some controversy over whether any preoperative biopsy is necessary or whether that simply represents an extra risk for the patient. This is a challenging subject because other disease entities enter into the differential diagnosis of a large abdominal mass involving the small intestine and mesentery. In general, clinical practice guidelines have suggested that resection may be performed without antecedent biopsy if a GIST is strongly suspected and if surgery can be accomplished without significant risk of morbidity to the patient. If only radical surgery leading to significant functional impairment could remove the lesion, it may be in the patient’s best interest to consider a preoperative biopsy to establish the diagnosis of a GIST. The prognosis of GISTs involving the small intestine is related to the adequacy of resection.⁶^,⁴⁵ In one series of 50 GIST patients involving the small intestine, 70% could be completely resected, with median overall survival longer than five years for patients with localized or locally advanced disease. Patients who underwent complete resections exhibited a five-year overall survival rate of 42%, while those whose lesions could not be completely resected had a five-year overall survival rate of only 8%.²⁸

Colorectal Tumors

GIST lesions are rare in the colon and rectum, accounting for approximately 5% of GIST cases. Perirectal and rectal GISTs present unique management challenges. As in other sites, small GIST lesions in the rectum may present as small, hard nodules less than 1 cm in diameter found incidentally during a clinical examination. However, much larger tumors can ulcerate, perhaps mimicking a rectal adenocarcinoma. Diagnosis can be challenging because the epithelioid or mixed cell variants of GISTs can also be misclassified as adenocarcinomas, especially if small biopsies are confounded by severe inflammatory changes or associated abscess formation. GIST lesions in this location present similar challenges to those in the small intestine; specifically, if nonmutilating surgery can achieve negative margins for a tiny GIST (<1 cm), that is probably a reasonable way to proceed. However, many GIST lesions in the rectal region will prove more challenging to resect, and surgery may only be feasible with significant functional morbidity. In such cases, it may be best to consider the patient unresectable without unacceptable risk and to consider preoperative therapy with imatinib to decrease the tumor bulk, followed by complete resection after maximal tumor shrinkage has occurred. In limited data available from older series, recurrence of rectal GIST has been common, with relatively poor prognosis. It remains to be seen whether newer approaches using multidisciplinary therapy, including imatinib, might improve these outcomes for patients with better functional results.

IMAGING STUDIES

Endoscopic Ultrasonography

Endoscopic ultrasonography (EUS) is a useful technology for evaluating possible GIST lesions because of their submucosal localization. The paradigm of a GIST lesion visualized by EUS is that of a hypoechoic mass contiguous with the fourth (muscularis propria) or second (muscularis mucosae) layers of the normal gut wall. In one study,⁴⁶ the features most predictive of so-called benign GI tumors were regular margins, tumor size 3 cm or smaller, and a homogeneous echogenicity pattern. Multivariant analysis identified the presence of cystic spaces and irregular margins as independent predictors of malignant potential. A second study identified tumor size larger than 4 cm, irregular extraluminal borders, echogenic foci larger than 3 mm, and cystic spaces larger than 4 mm as factors that correlated with malignant behavior in GIST.⁴⁷

Computed Tomography and Magnetic Resonance Imaging

CT is the most effective way to image primary lesions in the stomach because the oral contrast will outline masses and gastric thickening (Fig. 30-4). Differential diagnosis with inadequate gastric distention can be a challenge, especially when monitoring for recurrence following surgery. CT is also essential to stage the extent of disease completely and accurately. For measurable GIST lesions, it is particularly useful to perform CT with noncontrast image acquisition, as well as assess early and late images following the administration of intravenous contrast.